g1ddlZddlZddlmZmZddlmZGddeZGddeZ Gd d e Z Gd d e Z Gd deZ GddeZ dS)N)DatasetIterableDataset) ModelOutputc eZdZdZdZdZdS)PipelineDatasetc0||_||_||_dSNdatasetprocessparams)selfr r rs [/var/www/html/ai-engine/env/lib/python3.11/site-packages/transformers/pipelines/pt_utils.py__init__zPipelineDataset.__init__ s   c*t|jSr lenr rs r__len__zPipelineDataset.__len__4<   rcF|j|}|j|fi|j}|Sr r )riitem processeds r __getitem__zPipelineDataset.__getitem__s.|A DL5555 rN)__name__ __module__ __qualname__rrrrrrrsA !!!rrc.eZdZddZdZdZdZdZdS)PipelineIteratorNcj||_||_||_|dkrd}||_d|_d|_dS)a Roughly equivalent to ``` for item in loader: yield infer(item, **params) ``` Arguments: loader (`torch.utils.data.DataLoader` or `Iterable`): The iterator that will be used to apply `infer` on. infer (any function): The function to apply of each element of `loader`. params (`dict`): The parameters passed to `infer` along with every item loader_batch_size (`int`, *optional*): If specified, the items of `loader` are supposed to come as batch, and are loader_batched here making it roughly behave as ``` for items in loader: for i in loader_batch_size: item = items[i] yield infer(item, **params) ```N)loaderinferrloader_batch_size_loader_batch_index_loader_batch_data)rr&r'rr(s rrzPipelineIterator.__init__sJ6    ! ! $ !2$( "&rc*t|jSr )rr&rs rrzPipelineIterator.__len__?s4;rc8t|j|_|Sr iterr&iteratorrs r__iter__zPipelineIterator.__iter__BT[))  rctjtjr'jjd}n)i}jD]\}}t|tr|}t|dtjrtfd|D||<n>t|dtj rtfd|D||<|dvrt|trt|dtjrtfd|D||<n>t|dtj rtfd|D||<E|d||<Nt|jtjr%|jd||<t|jtj r%tj |jd||<|j||<j |}xjdz c_|S) ze Return item located at `loader_batch_index` within the current `loader_batch_data`. rc3XK|]$}|jdV%dSrNr) unsqueeze.0elrs r z5PipelineIterator.loader_batch_item..U;1n1n`b"T5M2N2X2XYZ2[2[1n1n1n1n1n1nrc3XK|]$}tj|jdV%dSr4np expand_dimsr)r7s rr:z5PipelineIterator.loader_batch_item..W91t1tfh".DD\A]_`2a2a1t1t1t1t1t1tr> attentions hidden_statespast_key_valuesc3XK|]$}|jdV%dSr4r5r7s rr:z5PipelineIterator.loader_batch_item..\r;rc3XK|]$}tj|jdV%dSr4r=r7s rr:z5PipelineIterator.loader_batch_item..^r@rNr%) isinstancer*torchTensorr)r6itemsrto_tupletupler>ndarrayr? __class__)rresultloader_batchedkelements` rloader_batch_itemz"PipelineIterator.loader_batch_itemFs d-u| < <' G,T-EFPPQRSSFF N"5;;== J J 7g{33%..00G!'!*el;;u,11n1n1n1nfm1n1n1n,n,nq))#GAJ ;;u,11t1t1t1tls1t1t1t,t,tq)JJJzZachOiOiJ!'!*el;;u,11n1n1n1nfm1n1n1n,n,nq))#GAJ ;;u,11t1t1t1tls1t1t1t,t,tq)?(,N1%%(@ A5<PP J)00H(I(S(STU(V(VN1%%(@ A2:NNJ)+wt?W7XZ[(\(\N1%%)00H(IN1%%,66~FFF   A%   rc|j$|j|jkr|St|j}|j|fi|j}|jt|tj r|}nMt|tr |d}n/t| d}||}t|trt|}n |jd}d|cxkr |jkr nn||_t|tr|dn||_d|_|S|S)Nr)r)r(rRnextr/r'rrFrGrHrKlistkeysrshaper*)rrr first_tensorkeyobserved_batch_sizes r__next__zPipelineIterator.__next__usd  # /D4LtOe4e4e))++ +DM""DJt33t{33  ! -)U\22 .( Iu-- .(| 9>>++,,Q/(~ ,-- <&),&7&7##&2&8&;#&????)??????*=&6@E6R6R&aillXaD #'(D $))++ + rr )rrr rrr0rRr[r!rrr#r#sf%'%'%'%'N   ---^"""""rr#c,eZdZdfd ZdZdZxZS)PipelineChunkIteratorNcNt|||dS)aI Roughly equivalent to ``` for iterator in loader: for item in iterator: yield infer(item, **params) ``` Arguments: loader (`torch.utils.data.DataLoader` or `Iterable`): The iterator that will be used to apply `infer` on. infer (any function): The function to apply of each element of `loader`. params (`dict`): The parameters passed to `infer` along with every item N)superr)rr&r'rr(rMs rrzPipelineChunkIterator.__init__s'$ /////rcFt|j|_d|_|Sr )r.r&r/ subiteratorrs rr0zPipelineChunkIterator.__iter__s T[))  rc2|j+ |jt|jfi|j|_ t|j}nN#t $rA|jt|jfi|j|_t|j}YnwxYw|Sr )rar'rTr/r StopIteration)rrs rr[zPipelineChunkIterator.__next__s   # \)tz$t}*=*=MMMMD  /T-..II / / / *tz$t}*=*=MMMMD T-..III /sA ABBr )rrr rr0r[ __classcell__)rMs@rr]r]s[000000( rr]ceZdZdZdZdZdS)PipelinePackIteratora Roughly equivalent to ``` packed = [] for item in loader: packed.append(item) if item["is_last"]: yield packed packed = [] ``` but it also handles cases where `item` are batched (meaning it's a dict of Tensor with first dimension > 1. In that case it does ``` packed = [] for batch in loader: # item is batched for item in batch: packed.append(item) if item["is_last"]: yield packed packed = [] ``` Arguments: loader (`torch.utils.data.DataLoader` or `Iterable`): The iterator that will be used to apply `infer` on. infer (any function): The function to apply of each element of `loader`. params (`dict`): The parameters passed to `infer` along with every item loader_batch_size (`int`, *optional*): If specified, the items of `loader` are supposed to come as batch, and are loader_batched here making it roughly behave as ``` for items in loader: for i in loader_batch_size: item = items[i] yield infer(item, **params) ```c8t|j|_|Sr r-rs rr0zPipelinePackIterator.__iter__r1rcd}g}|jr|j|jkrb|j|jkrR|}|d}|||r|S|j|jkR|sg|jt |jfi|j}|j t|tj r|}n/t| d}||}t|trt|}n |jd}d|cxkr |jkr nn||_||_d|_|j|jkrR|}|d}|||r|S|j|jkRn,|}|d}|||g|S)NFis_lastr)r)r(rRpopappendr'rTr/rrFrGrHrUrVrrWr*)rri accumulatorrrrXrYrZs rr[zPipelinePackIterator.__next__s!  # /D4LtOe4e4e*T-CCC--//((9--""4((('&& *T-CCC )" 4 #6#6FF$+FFI%1i662#,LLy~~//003C#,S>LlD11@*-l*;*;''*6*rGtorch.utils.datarr utils.genericrrr#r]rfrorxr!rrrsi 55555555''''''     g   @@@@@@@@F*****,***Zaaaaa+aaaH ) ) ) ) ) ) ) ) ] ] ] ] ]W ] ] ] ] ]r